If I'll have a heap dump, how can you find out what are the types of the nodes that are saved there? This is an example of a histogram I took in a different run: 1: 32272719 774545256 org.opendaylight.yangtools.yang.data.impl.schema.builder.impl.ImmutableLeafNodeBuilder$ImmutableLeafNode 2: 24618033 590832792 org.opendaylight.yangtools.util.SharedSingletonMap$Unordered 3: 15800515 512288528 [Ljava.lang.Object; 4: 16921694 406120656 org.opendaylight.yangtools.yang.data.impl.schema.builder.impl.ImmutableContainerNodeBuilder$ImmutableContainerNode 5: 15114335 362744040 org.opendaylight.yangtools.util.ImmutableOffsetMap$Unordered 6: 2565625 306451360 [C 7: 7655438 244974016 org.opendaylight.yangtools.yang.data.api.YangInstanceIdentifier$NodeIdentifierWithPredicates 8: 5640289 225611560 java.math.BigInteger 9: 7645644 183495456 org.opendaylight.yangtools.yang.data.impl.schema.builder.impl.ImmutableMapEntryNodeBuilder$ImmutableMapEntryNode 10: 7492069 179809656 org.opendaylight.yangtools.yang.data.impl.schema.builder.impl.ImmutableChoiceNodeBuilder$ImmutableChoiceNode
The types are already normalized to YANG Nodes instead of Flow,Interface,etc. So if I'll have an actual heap dump, how can I extract from that the real entities? -----Original Message----- From: Robert Varga [mailto:n...@hq.sk] Sent: Wednesday, January 11, 2017 2:01 AM To: Sela, Guy <guy.s...@hpe.com>; Tom Pantelis <tompante...@gmail.com> Cc: controller-dev@lists.opendaylight.org; odl netvirt dev <netvirt-...@lists.opendaylight.org> Subject: Re: [controller-dev] [mdsal-dev] Netvirt Scale tests: OutOfMemory from datastore On 01/10/2017 10:08 PM, Sela, Guy wrote: > It was a netvirt scale scenario. > I had one openstack Control with 100 Computes, i.e., 100 OVSs. > Had 1 private network and started creating VMs on each OVS. > This causes a mesh of tunnels between all OVSs, which means I got to 10000 > interfaces just for the tunnels, and a very big transport zone. > I had a lot of flows too obviously. Ah, the joys of a full mesh network :) I think PCEP scaled to 64K tunnels without too much trouble way back in Lithium, but then it does not need to deal with flows. Anyway, it would good to analyze the heap dump to see if there are any dedup opportunities way may be missing -- won't help with the issue at hand, but may lower footprint/increase throughput in future. Regards, Robert _______________________________________________ controller-dev mailing list controller-dev@lists.opendaylight.org https://lists.opendaylight.org/mailman/listinfo/controller-dev
